A Generative Adversarial based Approach for Continual Federated Learning with Non-IID Data
Bibliographic record
Abstract
Federated learning trains a shared model across many clients without moving raw data, while continual learning learns a stream of tasks and mitigates catastrophic forgetting. Continual federated learning combines these goals but is challenged by non-IID label skew and forgetting under evolving client data. We propose GAN-CFL (Generative Adversarial Networks-based Continual Federated Learning) to tackle these challenges. GAN-CFL enables clients to learn from new data, without storing historical data, and effectively adapts to non-IID data distributions. GAN-CFL enhances data heterogeneity by generating synthetic data to augment real datasets and mitigates catastrophic forgetting across multiple clients by incorporating elastic weight consolidation algorithm. In this framework, the generator produces synthetic images, while the discriminator classifies both real and generated images. The trained discriminator is then used as a classifier on real data to provide accuracy metrics. The global model aggregates local weights from clients to optimize overall performance. We evaluate GAN-CFL on six datasets, MNIST, K-MNIST, Fashion-MNIST, EMNIST-letters, EMNIST-Balanced, and CIFAR-10 for experiments involving up to 100 clients. Our model is compared to a centralized learning method for ablation analysis. The results show that GAN-CFL outperforms existing methods in classification accuracy.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.006 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.003 | 0.002 |
| Research integrity | 0.001 | 0.003 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".